Office CMBS delinquency at 12.2% is the headline, but the more important number in September's data is 4.435%. That is the fixed coupon on the $1.1 billion Hollywood loan that defaulted at maturity and was then extended for 14 months, at a rate far below anything the market would offer today. The extension came with conditions: the sponsors put in fresh cash, and the lender took every dollar of excess property cash flow. That is a lender selling time. When a special servicer starts charging for time, the decision has moved from "wait for rates" to "show us the tenants," and the cost has moved from bondholders to sponsor equity. Credit markets have already made that switch. Equity commentary that still treats delinquency as a stock of bad loans waiting for a rate cut has not.
The Modification Is the Signal
Look at the structure of the Hollywood deal. The loan is backed by five Class A office towers built between 2008 and 2021 and three film-studio properties, 2.2 million square feet within a mile of each other. Blackstone Property Partners and Hudson Pacific Properties are the borrowers. It was securitized in 2021, moved to special servicing in July 2026, matured in August, and was not repaid.
SitusAMC is the special servicer. It is the largest in the business by unpaid principal balance, at $111 billion as of January. It did not foreclose. It granted an extension to November 9, 2027, kept the 4.435% coupon, and set three conditions. The borrowers fund a $20 million leasing reserve from money outside the property. A full cash trap stays in place until the loan is repaid. All excess cash flow goes into the reserve.
Each condition tells you something. The $20 million from outside sources means the sponsors had to write a check to keep their option alive. The cash trap means this collateral sends nothing upstream to the equity until the debt is gone. The 14-month term is short, which shows the servicer is not betting on lower rates. It is betting on one event: whether Netflix and 20th Century Fox renew.
The concentration is severe. Netflix leases 57.8% of the space. One Netflix lease covering 7.0% of the space expires in January 2027, and another covering 17.2% expires in June 2028. Fox's lease, 6.4% of the space, expires in December 2026. Fitch, citing the servicer, said Fox's renewal depends on whether its studio productions are picked up for more seasons. Occupancy already fell from 91% in September 2025 to 84.9% in July. If the 30% of space that is rolling does not renew, occupancy falls to roughly 55%.
So the credit question on a $1.1 billion bond deal comes down to television renewal decisions. Fitch has moved the outlook to negative and has tied a downgrade to the borrower's ability to address that rollover. The ratings agency is watching the leasing calendar, not the Fed.
Fixed Coupons, Floating Pain: Hollywood Versus Houston
The second-largest driver of September's increase shows the other half of the problem. A $470 million interest-only loan on One Allen Center and Three Allen Center in downtown Houston was also securitized in 2021. The towers were completed in 1972 and 1980, total about 2.2 million square feet with a parking garage, and are owned by Brookfield Properties. The loan floats at SOFR plus 3.08%.
When the loan was written, SOFR was near zero, so the all-in coupon was about 3.1%. With SOFR around 3.88% after the Fed's most recent hike, the coupon is about 6.96%. One more hike would push it to about 7.2%. Interest cost on an interest-only loan has more than doubled, and the buildings are 71% occupied. S&P flagged in March that the loan might not be repaid at its final extended maturity in April. It was not. Trepp suggests the September move to non-performing status may reflect ongoing negotiations.
Together, the two loans show that office distress comes through two separate channels.
- Fixed-rate loans fail at maturity. The coupon is fine, but no lender will refinance at today's rates against today's rent roll. Hollywood pays 4.435% and still cannot be taken out.
- Floating-rate loans fail during the term. Debt service rises with SOFR until property income cannot cover it. Houston became unaffordable before it reached a final maturity.
Both channels lead to the same place: the gap between what the asset was financed at and what it can be financed at today. Bisnow's comparison in the source material shows the size of that gap. The 10-year Treasury closed at 1.52% on September 30, 2021, and at 5.29% five years later to the day. By October 2 it was 5.28%, with the 2-year at 4.83%. The curve is positively sloped by 45 basis points, so even a fall in short rates leaves the long-end refinancing benchmark high.
Why 12.2% Understates the Stress
The delinquency rate measures loans that are failing right now. It does not count loans that have already been fixed on paper.
The Hollywood loan shows why. It pushed September's rate up as a maturity default. After the modification, with a new maturity date and a coupon the borrower can pay, it will typically be reported as current again, though that depends on how each tracker classifies modified loans. The same collateral, the same rollover risk, and a loan on negative outlook would then move out of the delinquency count and into the performing count. The headline number improves while the underlying risk stays the same.
So a 12.2% reading, 1.5 points above the worst of the Financial Crisis and just below January 2026's 12.3% record, is a floor on office credit stress, not the full measure. Every extension pushes a default into a later month. Every cash trap delays the loss for bondholders and makes it certain for the equity. The rate peaked in January, eased, and spiked again because the pool of loans still waiting for resolution keeps producing new maturity defaults faster than extensions can absorb them.
The source makes another point quietly: the originating banks are off the hook. These loans were split into slices and sold to institutional investors worldwide in 2021. Losses do not show up as bank charge-offs. They show up as appraisal reductions, interest shortfalls on junior bonds, and eventually write-downs spread across insurers, pensions, and credit funds. That is why office CMBS can pass the 2008 peak without a banking crisis. The damage is spread out and slow to appear.
The Case That Extend-and-Pretend Still Works
There is a serious argument that the servicers are right and the alarm is overdone.
First, the trend in the headline rate is flat, not rising. September's 12.2% is below January's 12.3%. A market heading toward collapse does not usually hold steady for nine months.
Second, the Hollywood collateral is not obsolete. The towers are Class A and built between 2008 and 2021. The tenant is Netflix, which has been there long enough to occupy more than half the space. Lease discussions with Netflix are ongoing, according to SitusAMC as cited by Fitch. If Netflix renews the June 2028 lease and the Fox productions continue, the cash trap fills the reserve, the loan can be refinanced on a stronger rent roll, and extending it will have been the right call. Foreclosing on a property that is 85% leased to a major tenant would have locked in a loss the servicer did not need to take.
Third, the sponsors put in money. Borrowers who see no value in an asset do not add $20 million of outside capital to keep it. Blackstone and Hudson Pacific clearly believe the equity is worth more than zero.
All of this is reasonable, and it actually supports the thesis. The extension works only if the tenants renew. It does not work just because time passes. The industry slogans moved from "survive till '25" to "bliss in '26" to "heaven in '27," and each one assumed lower rates would solve the problem. The Hollywood terms assume nothing of the kind. They keep a below-market coupon only for a short window, collect cash in the meantime, and push the outcome to a date close to the leasing decisions. Houston, an older building at 71% occupancy on a floating rate tied to a Fed that is still hiking, does not have the same tenant-driven path out. That is why it is still in negotiations rather than modified.
Bisnow's line, "sell at today's price, or the lender will," is the conclusion credit has already reached. Extensions now go to loans with a believable leasing path. Loans without one are being moved toward sale.
What Ends the Pretending
Three dates and one disclosure will test this read.
The first is December 2026, when the Fox lease expires. That is the earliest point where the Hollywood loan's leasing path either holds up or breaks. The second is January 2027, when the smaller Netflix lease expires and shows whether the anchor tenant is renewing, shrinking, or leaving. The third is November 9, 2027, the extended maturity, which comes seven months before the larger Netflix lease expires. Unless Netflix commits early, the servicer will face the same decision next year with less occupancy behind it.
The disclosure to watch is the Houston resolution. If Brookfield gets a Hollywood-style extension with fresh equity and a cash trap, servicers are still paying to wait, even for older buildings. If it ends in a discounted sale, a deed-in-lieu, or a note sale, then "sell at today's price" has become the actual policy. The loss severity would then give the market a real price for 1970s and 1980s downtown office space, which no appraisal has yet provided.
The view changes if delinquency falls because maturing loans are refinanced with third-party capital, not because they are modified. That would mean the refinancing gap is closing. With the 10-year at 5.28% and two of September's largest defaults financed when rates were near zero, nothing in the current data shows that happening. Until it does, the falling delinquency rate after each modification should be read as distress being postponed, not removed.
